So Let me ask a question. As a Network Architect when are we going to see the cloud vendors begin offering Network services in the same tiered manner as Storage? What I mean is have flavors of networking with varying SLA/redundancy/connection throughput, at different cost. This would mean doing the same that the storage teams have done and build different infrastructure on different vendors and at different cost. It would also require building billing/costing models to support this function, but reusing the Storage Tiering model seems a good first start. It would also help educate Cloud users and Consumers that Infrastructure only matters for capability(SLA) and cost and not for model/serial number, port count, etc. I personally am tired of having Executive level discussions with clients wanting cloud who then want the serial numbers of all switches and to know how many free ports there are in the network.
The Question they should be asking are:

How much throughput do I have? And can I burst to what?

What level of redundancy do I have? Can you Guarantee a minimum outage regardless of hardware failure in the cloud?

What is the cost for the current throughput/capability/SLA?

#seangmuller